AI Email Tools

AI Email Tools — independent reviews, comparisons, pricing and step-by-step guides on Aizhi.

ChatGPT

ChatGPT is a generative artificial intelligence chatbot developed by OpenAI. Originally released in November 2022, the product uses large language models—specifically generative pre-trained transformers (GPTs)—to generate text, speech, and images in response to user prompts. ChatGPT accelerated the AI boom, an ongoing period marked by rapid investment and public attention toward the field of artificial intelligence (AI). OpenAI operates the service on a freemium model. Users can interact with ChatGPT through text, audio, and image prompts. ChatGPT was quickly adopted, reaching 100 million monthly active users two months after its release and 900 million weekly active users in February 2026. It has been lauded for its potential to transform numerous professional fields, and has instigated public debate about the nature of creativity and the future of knowledge work. The chatbot has also been criticized for its limitations and potential for unethical use. It can generate plausible-sounding but incorrect or nonsensical answers, known as hallucinations. Biases in its training data have been reflected in its responses. The chatbot can facilitate academic dishonesty, generate misinformation, and create malicious code. The ethics of its development, particularly the use of copyrighted content as training data, have also drawn controversy. == Features == ChatGPT is a chatbot and AI assistant built on large language model (LLM) technology. It is designed to generate human-like text and can carry out a wide variety of tasks. These include, among many others, writing and debugging computer programs, composing music, scripts, fairy tales, and essays, answering questions (sometimes at a level exceeding that of an average human test-taker), and generating business concepts. ChatGPT is frequently used for translation and summarization tasks, and can simulate interactive environments such as a Linux terminal, a multi-user chat room, or simple text-based games such as tic-tac-toe. Users interact with ChatGPT through conversations which consist of text, audio, and image inputs and outputs. The user's inputs to these conversations are referred to as prompts. An optional "Memory" feature allows users to tell ChatGPT to memorize specific information. Another option allows ChatGPT to recall old conversations. GPT-based moderation classifiers are used to reduce the risk of harmful outputs being presented to users. In March 2023, OpenAI added support for plugins for ChatGPT. This includes both plugins made by OpenAI, such as web browsing and code interpretation, and external plugins from developers such as Expedia, OpenTable, and Zapier. From October to December 2024, ChatGPT Search was deployed. It allows ChatGPT to search the web in an attempt to make more accurate and up-to-date responses. It increased OpenAI's direct competition with major search engines. OpenAI allows businesses to tailor how their content appears in the ChatGPT Search results and influence what sources are used. In December 2024, OpenAI launched a new feature allowing users to call ChatGPT with a telephone for up to 15 minutes per month for free. In September 2025, OpenAI added a feature called Pulse, which generates a daily analysis of a user's chats and connected apps such as Gmail and Google Calendar. In October 2025, OpenAI launched ChatGPT Atlas, a browser integrating the ChatGPT assistant directly into web navigation, to compete with existing browsers such as Google Chrome. It has an additional feature called "agentic mode" that allows it to take online actions for the user. === Paid tier === ChatGPT was initially free to the public and remains free in a limited capacity. In February 2023, OpenAI launched a premium service, ChatGPT Plus, that costs US$20 per month. What was offered on the paid plan versus the free tier changed as OpenAI has continued to update ChatGPT, and a Pro tier at $200/mo was introduced in December 2024. The Pro launch coincided with the release of the o1 model. In August 2025, ChatGPT Go was offered in India for ₹399 per month. The plan has higher limits than the free version. === Mobile apps === In May-July 2023, OpenAI began offering ChatGPT iOS and Android apps. ChatGPT can also power Android's assistant. An app for Windows launched on the Microsoft Store on October 15, 2024. === Languages === OpenAI met Icelandic President Guðni Th. Jóhannesson in 2022. In 2023, OpenAI worked with a team of 40 Icelandic volunteers to fine-tune ChatGPT's Icelandic conversation skills as a part of Iceland's attempts to preserve the Icelandic language. ChatGPT (based on GPT-4) was better able to translate Japanese to English when compared to Bing, Bard, and DeepL Translator in 2023. In December 2023, the Albanian government decided to use ChatGPT for the rapid translation of European Union documents and the analysis of required changes needed for Albania's accession to the EU. Several studies have shown that ChatGPT can outperform Google Translate in some mainstream translation tasks. However, as of 2024, no machine translation services match human expert performance. In August 2024, a representative of the Asia Pacific wing of OpenAI made a visit to Taiwan, during which a demonstration of ChatGPT's Chinese abilities was made. ChatGPT's Mandarin Chinese abilities were lauded, but the ability of the AI to produce content in Mandarin Chinese in a Taiwanese accent was found to be "less than ideal" due to differences between mainland Mandarin Chinese and Taiwanese Mandarin. === GPT Store === In November 2023, OpenAI released GPT Builder, a tool allowing users to customize ChatGPT's behavior for a specific use case. The customized systems are referred to as GPTs. In January 2024, OpenAI launched the GPT Store, a marketplace for GPTs. At launch, OpenAI included more than 3 million GPTs created by GPT Builder users in the GPT Store. === ChatGPT Apps === In September 2025, OpenAI added support for Model Context Protocol (MCP) to ChatGPT apps. When enabled in developer mode, this allows for improved third-party access to ChatGPT tools and servers. === Deep Research === In February 2025, OpenAI released Deep Research, a feature that generates reports based on extensive web searches. It was initially based on the reasoning model o3 and took 5 to 30 minutes per report. === Images === In October 2023, OpenAI's image generation model DALL-E 3 was integrated into ChatGPT. The integration used ChatGPT to write prompts for DALL-E guided by conversations with users. In March 2025, OpenAI updated ChatGPT to generate images using GPT Image instead of DALL-E. One of the most significant improvements was in the generation of text within images, which is especially useful for branded content. However, this ability is noticeably worse in non-Latin alphabets. The model can also generate new images based on existing ones provided in the prompt. These images are generated with C2PA metadata, which can be used to verify that they are AI-generated. OpenAI has emplaced additional safeguards to prevent what the company deems to be harmful image generation. === Agents === In 2025, OpenAI added several features to make ChatGPT more agentic (capable of autonomously performing longer tasks). In January, Operator was released. It was capable of autonomously performing tasks through web browser interactions, including filling forms, placing online orders, scheduling appointments, and other browser-based tasks. It was controlling a software environment inside a virtual machine with limited internet connectivity and with safety restrictions. It struggled with complex user interfaces. In May 2025, OpenAI introduced an agent for coding named Codex. It is capable of writing software, answering codebase questions, running tests, and proposing pull requests. It is based on a fine-tuned version of OpenAI o3. It has two versions, one running in a virtual machine in the cloud, and one where the agent runs in the cloud, but performs actions on a local machine connected via API. In July 2025, OpenAI released ChatGPT agent, an AI agent that can perform multi-step tasks. Like Operator, it controls a virtual computer. It also inherits from Deep Research's ability to gather and summarize significant volumes of information. The user can interrupt tasks or provide additional instructions as needed. In September 2025, OpenAI partnered with Stripe, Inc. to release Agentic Commerce Protocol, enabling purchases through ChatGPT. At launch, the feature was limited to purchases on Etsy from US users with a payment method linked to their OpenAI account. OpenAI takes an undisclosed cut from the merchant's payment. === ChatGPT Health === On January 7, 2026, OpenAI introduced a feature called "ChatGPT Health", whereby ChatGPT can discuss the user's health in a way that is separate from other chats. The feature is not available for users in the United Kingdom, Switzerland, or the European Economic Area, and is available on a waitli
Read more →
Data transformation (computing)

In computing, data transformation is the process of converting data from one format or structure into another format or structure. It is a fundamental aspect of most data integration and data management tasks such as data wrangling, data warehousing, data integration and application integration. Data transformation can be simple or complex based on the required changes to the data between the source (initial) data and the target (final) data. Data transformation is typically performed via a mixture of manual and automated steps. Tools and technologies used for data transformation can vary widely based on the format, structure, complexity, and volume of the data being transformed. A master data recast is another form of data transformation where the entire database of data values is transformed or recast without extracting the data from the database. All data in a well-designed database is directly or indirectly related to a limited set of master database tables by a network of foreign key constraints. Each foreign key constraint is dependent upon a unique database index from the parent database table. Therefore, when the proper master database table is recast with a different unique index, the directly and indirectly related data are also recast or restated. The directly and indirectly related data may also still be viewed in the original form since the original unique index still exists with the master data. Also, the database recast must be done in such a way as to not impact the applications architecture software. When the data mapping is indirect via a mediating data model, the process is also called data mediation. == Data transformation process == Data transformation can be divided into the following steps, each applicable as needed based on the complexity of the transformation required. Data discovery Data mapping Code generation Code execution Data review These steps are often the focus of developers or technical data analysts who may use multiple specialized tools to perform their tasks. The steps can be described as follows: Data discovery is the first step in the data transformation process. Typically the data is profiled using profiling tools or sometimes using manually written profiling scripts to better understand the structure and characteristics of the data and decide how it needs to be transformed. Data mapping is the process of defining how individual fields are mapped, modified, joined, filtered, aggregated etc. to produce the final desired output. Developers or technical data analysts traditionally perform data mapping since they work in the specific technologies to define the transformation rules (e.g. visual ETL tools, transformation languages). Code generation is the process of generating executable code (e.g. SQL, Python, R, or other executable instructions) that will transform the data based on the desired and defined data mapping rules. Typically, the data transformation technologies generate this code based on the definitions or metadata defined by the developers. Code execution is the step whereby the generated code is executed against the data to create the desired output. The executed code may be tightly integrated into the transformation tool, or it may require separate steps by the developer to manually execute the generated code. Data review is the final step in the process, which focuses on ensuring the output data meets the transformation requirements. It is typically the business user or final end-user of the data that performs this step. Any anomalies or errors in the data that are found and communicated back to the developer or data analyst as new requirements to be implemented in the transformation process. == Types of data transformation == === Batch data transformation === Traditionally, data transformation has been a bulk or batch process, whereby developers write code or implement transformation rules in a data integration tool, and then execute that code or those rules on large volumes of data. This process can follow the linear set of steps as described in the data transformation process above. Batch data transformation is the cornerstone of virtually all data integration technologies such as data warehousing, data migration and application integration. When data must be transformed and delivered with low latency, the term "microbatch" is often used. This refers to small batches of data (e.g. a small number of rows or a small set of data objects) that can be processed very quickly and delivered to the target system when needed. === Benefits of batch data transformation === Traditional data transformation processes have served companies well for decades. The various tools and technologies (data profiling, data visualization, data cleansing, data integration etc.) have matured and most (if not all) enterprises transform enormous volumes of data that feed internal and external applications, data warehouses and other data stores. === Limitations of traditional data transformation === This traditional process also has limitations that hamper its overall efficiency and effectiveness. The people who need to use the data (e.g. business users) do not play a direct role in the data transformation process. Typically, users hand over the data transformation task to developers who have the necessary coding or technical skills to define the transformations and execute them on the data. This process leaves the bulk of the work of defining the required transformations to the developer, which often in turn do not have the same domain knowledge as the business user. The developer interprets the business user requirements and implements the related code/logic. This has the potential of introducing errors into the process (through misinterpreted requirements), and also increases the time to arrive at a solution. This problem has given rise to the need for agility and self-service in data integration (i.e. empowering the user of the data and enabling them to transform the data themselves interactively). There are companies that provide self-service data transformation tools. They are aiming to efficiently analyze, map and transform large volumes of data without the technical knowledge and process complexity that currently exists. While these companies use traditional batch transformation, their tools enable more interactivity for users through visual platforms and easily repeated scripts. Still, there might be some compatibility issues (e.g. new data sources like IoT may not work correctly with older tools) and compliance limitations due to the difference in data governance, preparation and audit practices. === Interactive data transformation === Interactive data transformation (IDT) is an emerging capability that allows business analysts and business users the ability to directly interact with large datasets through a visual interface, understand the characteristics of the data (via automated data profiling or visualization), and change or correct the data through simple interactions such as clicking or selecting certain elements of the data. Although interactive data transformation follows the same data integration process steps as batch data integration, the key difference is that the steps are not necessarily followed in a linear fashion and typically don't require significant technical skills for completion. There are a number of companies that provide interactive data transformation tools, including Trifacta, Alteryx and Paxata. They are aiming to efficiently analyze, map and transform large volumes of data while at the same time abstracting away some of the technical complexity and processes which take place under the hood. Interactive data transformation solutions provide an integrated visual interface that combines the previously disparate steps of data analysis, data mapping and code generation/execution and data inspection. That is, if changes are made at one step (like for example renaming), the software automatically updates the preceding or following steps accordingly. Interfaces for interactive data transformation incorporate visualizations to show the user patterns and anomalies in the data so they can identify erroneous or outlying values. Once they've finished transforming the data, the system can generate executable code/logic, which can be executed or applied to subsequent similar data sets. By removing the developer from the process, interactive data transformation systems shorten the time needed to prepare and transform the data, eliminate costly errors in the interpretation of user requirements and empower business users and analysts to control their data and interact with it as needed. == Transformational languages == There are numerous languages available for performing data transformation. Many transformation languages require a grammar to be provided. In many cases, the grammar is structured using something closely resembling Backus–Naur form (BNF). There are numerous languages
Read more →
Content-oriented workflow models

In data management, a content-oriented workflow model seeks to articulate workflow progression by the presence of content units (like data-records/objects/documents). Most content-oriented workflow approaches provide a life-cycle model for content units, such that workflow progression can be qualified by conditions on the state of the units. Most approaches are research and work in progress and the content models and life-cycle models are more or less formalized. The term content-oriented workflows is an umbrella term for several scientific workflow approaches, namely "data-driven", "resource-driven", "artifact-centric", "object-aware", and "document-oriented". Thus, the meaning of "content" ranges from simple data attributes to self-contained documents; the term "content-oriented workflows" appeared at first in as an umbrella term. Such a general term, independent from a specific approach, is necessary to contrast the content-oriented modelling principle with traditional activity-oriented workflow models (like Petri nets or BPMN) where a workflow is driven by a control flow and where the content production perspective is neglected or even missing. The term "content" was chosen to subsume the different levels in granularity of the content units in the respective workflow models; it was also chosen to make associations with content management. Both terms "artifact-centric" and "data-driven" would also be good candidates for an umbrella term, but each is closely related to a specific approach of a single working group. The "artifact-centric" group itself (i.e. IBM Research) has generalized the characteristics of their approach and has used "information-centric" as an umbrella term in. Yet, the term information is too unspecific in the context of computer science, thus, "content-orientated workflows" is considered as good compromise. == Workflow Model Approaches == === Data-driven === The data-driven process structures provides a sophisticated workflow model being specialized on hierarchical write-and-review-processes. The approach provides interleaved synchronization of sub-processes and extends activity diagrams. Unfortunately, the COREPRO prototype implementation is not publicly available. Research on the project had been ceased. The general idea has been continued by Reichert in form of the #Object-aware approach. Synonyms data-driven process structures / data-driven modeling and coordination Protagonists Dr. Dominic Müller (University of Twente), Joachim Herbst (DaimlerChrysler Research), and Manfred Reichert (at this time Assoc. Prof. at Univ. of Twente, currently Prof. at Ulm Univ.) Organization(s) University of Twente, DaimlerChrysler Period 2005 - 2007 Selected publications Implementation COREPRO === Resource-driven === The resource-driven workflow system is an early approach that considered workflows from a content-oriented perspective and emphasizes on the missing support for plain document-driven processes by traditional activity-oriented workflow engines. The resource-driven approach demonstrated the application of database triggers for handling workflow events. Still the system implementation is centralized and the workflow schema is statically defined. The project appeared in 2005 but many aspects are considered future work by the authors. Research did not continue on the project. Wang completed his PhD thesis in 2009, yet, his thesis does not mention the resource-driven approach to workflow modelling but is about discrete event simulation. Synonyms Resource-based Workflows / Document-Driven Workflow Systems Protagonists Jianrui Wang and Prof. Akhil Kumar Organization Pennsylvania State University Period 2005 - today Selected publications Implementation N/A === Artifact-centric === The artifact-centric approach provides a framework for content-oriented workflows. In this model, the enterprise application landscape includes distributed business services, while the workflow engine is centralized. Process enactment is integrated with database management system infrastructure, and the project is funded by IBM. Synonyms artifact-centric business process models / artifact-based business process (ACP) / artifact-centric workflows Protagonists Richard Hull and Dr. Kamal Bhattacharya as well as Cagdas E. Gerede and Jianwen Su Organization IBM (T.J. Watson Research Center, NY) Period 2007 - today Selected publications Implementation ArtiFact === Object-aware === The object-aware approach manages a set of object types and generates forms for creating object instances. The form completion flow is controlled by transitions between object configurations each describing a progressing set of mandatory attributes. Each object configuration is named by an object state. The data production flow is user-shifting and it is discrete by defining a sequence of object states. The discussion is currently limited to a centralized system, without any workflows across different organizations. However, the approach is of great relevance to many domains like concurrent engineering. Finally, the object-aware approach and its PHILharmonicFlows system are going to provide general-purpose workflow systems for generic enactment of data production processes. Synonyms object-aware process management / datenorientiertes Prozess-Management-System Protagonists Vera Künzle and Prof. Manfred Reichert Organization Ulm University Period 2009 - today Selected publications Implementation PHILharmonicFlows === Distributed Document-oriented === Distributed document-oriented process management (dDPM) enables distributed case handling in heterogeneous system environments and it is based on document-oriented integration. The workflow model reflects the paper-based working practice in inter-institutional healthcare scenarios. It targets distributed knowledge-driven ad hoc workflows, wherein distributed information systems are required to coordinate work with initially unknown sets of actors and activities. The distributed workflow engine supports process planning & process history as well as participant management and process template creation with import/export. The workflow engine embeds a functional fusion of 1) group-based instant messaging 2) with a shared work list editor 3) with version control. The software implementation of dDPM is α-Flow which is available as open source. dDPM and α-Flow provide a content-oriented approach to schema-less workflows. The complete distributed case handling application is provided in form of a single active Document ("α-Doc"). The α-Doc is a case file (as information carrier) with an embedded workflow engine (in form of active properties). Inviting process participants is equivalent to providing them with a copy of an α-Doc, copying it like an ordinary desktop file. All α-Docs that belong to the same case can synchronize each other, based on the participant management, electronic postboxes, store-and-forward messaging, and an offline-capable synchronization protocol. Synonyms distributed document-oriented process management (dDPM), distributed case handling via active documents Protagonists Christoph P. Neumann and Prof. Richard Lenz Organization Friedrich-Alexander-Universität Erlangen-Nürnberg Period 2009 - 2012 Selected Publications and a PhD thesis Implementation α-Flow (open source) == Related Concepts == === Content Management === The bandwidth of Content management systems (CMS) reaches from Web content management systems (WCMS) and Document management system (DMS) to Enterprise Content Management (ECM). Mature DMS products support document production workflows in a basic form, primarily focusing on review cycle workflows concerning a single document. === Groupware and Computer-Supported Cooperative Work === Groupware focuses on messaging (like E-Mail, Chat, and Instant Messaging), shared calendars (e.g. Lotus Notes, Microsoft Outlook with Exchange Server), and conferencing (e.g. Skype). Groupware overlaps with Computer-supported cooperative work (CSCW), that originated from shared multimedia editors (for live drawing/sketching) and synchronous multi-user applications like desktop sharing. The extensive conceptual claim of CSWC must be put into perspective by its actual solution scope, that is available as the CSCW Matrix. === Case Handling === The case handling paradigm stems from Prof. van der Aalst and gained momentum in 2005. The core features are: (a) provide all information available, i.e. present the case as a whole rather than showing bits and pieces, (b) decide about activities on the basis of the information available rather than the activities already executed, (c) separate work distribution from authorization and allow for additional types of roles, not just the execute role, and (d) allow workers to view and add/modify data before or after the corresponding activities have been executed. In healthcare, the flow of a patient between healthcare professionals is considered as a workflow - with activities that inc
Read more →
Cognos ReportNet

Cognos ReportNet (CRN) was a web-based software product for creating and managing ad hoc and custom-made reports. ReportNet was developed by the Ottawa-based company Cognos (formerly Cognos Incorporated), an IBM company. The web-based reporting tool was launched in September 2003. Since IBM's acquisition of Cognos, ReportNet has been renamed IBM Cognos ReportNet like all other Cognos products. ReportNet uses web services standards such as XML and Simple Object Access Protocol and also supports dynamic HTML and Java. ReportNet is compatible with multiple databases including Oracle, SAP, Teradata, Microsoft SQL server, DB2 and Sybase. The product provides interface in over 10 languages, has Web Services architecture to meet the needs of multi-national, diversified enterprises and helps reduce total cost of ownership. Multiple versions of Cognos ReportNet have since been released by the company. Cognos ReportNet was awarded the Software and Information Industry Association (SIIA) 2005 Codie awards for the "Best Business Intelligence or Knowledge Management Solution" category. CRN's capabilities have been further used in IBM Cognos 8 BI (2005), the latest reporting tool. CRN comes with its own software development kit (SDK). == Launch == Early adopters of Cognos ReportNet for their corporate reporting needs included Bear Stearns, BMW and Alfred Publishing. Around this same time of launch, Cognos competitor Business Objects released version 6.1 of its enterprise reporting tool. Cognos ReportNet has been successful since its launch, raising revenues in 2004 from licensing fees. == Controversy == Cognos rival Business Objects announced in 2005 that BusinessObjects XI significantly outperformed Cognos ReportNet in benchmark tests conducted by VeriTest, an independent software testing firm. The tests performed showed Cognos ReportNet performed poorly when processing styled reports, complex business reports and combination of both. The tests reported a massive 21 times higher report throughput for BusinessObjects XI than Cognos ReportNet at capacity loads. Cognos soon dismissed the claims by stating Business Objects dictated the environment and testing criteria and Cognos did not provide the software to participate in benchmark test. Cognos later performed their own test to demonstrate Cognos ReportNet capabilities. == Components == Cognos Report Studio – A Web-based product for creating complex professional looking reports. Cognos Query Studio - A Web-based product for creating ad-hoc reports. Cognos Framework Manager – A metadata modeling tool to create BI metadata for reporting and dashboard applications. Cognos Connection – Main portal used to access reports, schedule reports and perform administrator activities. == Versions == Cognos ReportNet 1.1 – Java EE-style professional web-based authoring tool. (base version) Cognos ReportNet IBM Special Edition – comes with an embedded version of IBM WebSphere as its application server and IBM DB2 as its data store. Cognos Linux – for Intel-based Linux platforms.
Read more →
Purged cross-validation

Purged cross-validation is a variant of k-fold cross-validation designed to prevent look-ahead bias in time series and other structured data, developed in 2017 by Marcos López de Prado at Guggenheim Partners and Cornell University. It is primarily used in financial machine learning to ensure the independence of training and testing samples when labels depend on future events. It provides an alternative to conventional cross-validation and walk-forward backtesting methods, which often yield overly optimistic performance estimates due to information leakage and overfitting. == Motivation == Standard cross-validation assumes that observations are independently and identically distributed (IID), which often does not hold in time series or financial datasets. If the label of a test sample overlaps in time with the features or labels in the training set, the result may be data leakage and overfitting. Purged cross-validation addresses this issue by removing overlapping observations and, optionally, adding a temporal buffer ("embargo") around the test set to further reduce the risk of leakage. The figure below illustrates standard 5 Fold Cross-Validation == Purging == Purging removes from the training set any observation whose timestamp falls within the time range of formation of a label in the test set. This can be the case for train set observations before and after the test set. Their removal ensures that the algorithm cannot learn during train time information that will be used to assess the performance of the algorithm. See the figure below for an illustration of purging. == Embargoing == Embargoing addresses a more subtle form of leakage: even if an observation does not directly overlap the test set, it may still be affected by test events due to market reaction lag or downstream dependencies. To guard against this, a percentage-based embargo is imposed after each test fold. For example, with a 5% embargo and 1000 observations, the 50 observations following each test fold are excluded from training. Unlike purging, embargoing can only occur after the test set. The figure below illustrates the application of embargo: == Applications == Purged and embargoed cross-validation has been useful in: Backtesting of trading strategies Validation of classifiers on labeled event-driven returns Any machine learning task with overlapping label horizons == Example == To illustrate the effect of purging and embargoing, consider the figures below. Both diagrams show the structure of 5-fold cross-validation over a 20-day period. In each row, blue squares indicate training samples and red squares denote test samples. Each label is defined based on the value of the next two observations, hence creating an overlap. If this overlap is left untreated, test set information leaks into the train set. The second figure applies the Purged CV procedure. Notice how purging removes overlapping observations from the training set and the embargo widens the gap between test and training data. This approach ensures that the evaluation more closely resembles a true out-of-sample test and reduces the risk of backtest overfitting. == Combinatorial Purged Cross-Validation == Walk-forward backtesting analysis, another common cross-validation technique in finance, preserves temporal order but evaluates the model on a single sequence of test sets. This leads to high variance in performance estimation, as results are contingent on a specific historical path. Combinatorial Purged Cross-Validation (CPCV) addresses this limitation by systematically constructing multiple train-test splits, purging overlapping samples, and enforcing an embargo period to prevent information leakage. The result is a distribution of out-of-sample performance estimates, enabling robust statistical inference and more realistic assessment of a model's predictive power. === Methodology === CPCV divides a time-series dataset into N sequential, non-overlapping groups. These groups preserve the temporal order of observations. Then, all combinations of k groups (where k < N) are selected as test sets, with the remaining N − k groups used for training. For each combination, the model is trained and evaluated under strict controls to prevent leakage. To eliminate potential contamination between training and test sets, CPCV introduces two additional mechanisms: Purging: Any training observations whose label horizon overlaps with the test period are excluded. This ensures that future information does not influence model training. Embargoing: After the end of each test period, a fixed number of observations (typically a small percentage) are removed from the training set. This prevents leakage due to delayed market reactions or auto-correlated features. Each data point appears in multiple test sets across different combinations. Because test groups are drawn combinatorially, this process produces multiple backtest "paths," each of which simulates a plausible market scenario. From these paths, practitioners can compute a distribution of performance statistics such as the Sharpe ratio, drawdown, or classification accuracy. === Formal definition === Let N be the number of sequential groups into which the dataset is divided, and let k be the number of groups selected as the test set for each split. Then: The number of unique train-test combinations is given by the binomial coefficient: ( N k ) {\displaystyle {\binom {N}{k}}} Each observation is used in k {\displaystyle k} test sets and contributes to φ [ N , k ] {\displaystyle \varphi [N,k]} unique backtest paths: φ [ N , k ] = k N ( N k ) {\displaystyle \varphi [N,k]={\frac {k}{N}}{\binom {N}{k}}} This yields a distribution of performance metrics rather than a single point estimate, making it possible to apply Monte Carlo-based or probabilistic techniques to assess model robustness. === Illustrative example === Consider the case where N = 6 and k = 2. The number of possible test set combinations is ( 6 2 ) = 15 {\displaystyle {\binom {6}{2}}=15} . Each of the six groups appears in five test splits. Consequently, five distinct backtest paths can be constructed, each incorporating one appearance from every group. ==== Test group assignment matrix ==== This table shows the 15 test combinations. An "x" indicates that the corresponding group is included in the test set for that split. ==== Backtest path assignment ==== Each group contributes to five different backtest paths. The number in each cell indicates the path to which the group's result is assigned for that split. === Advantages === Combinatorial Purged Cross-Validation offers several key benefits over conventional methods: It produces a distribution of performance metrics, enabling more rigorous statistical inference. The method systematically eliminates lookahead bias through purging and embargoing. By simulating multiple historical scenarios, it reduces the dependence on any single market regime or realization. It supports high-confidence comparisons between competing models or strategies. CPCV is commonly used in quantitative strategy research, especially for evaluating predictive models such as classifiers, regressors, and portfolio optimizers. It has been applied to estimate realistic Sharpe ratios, assess the risk of overfitting, and support the use of statistical tools such as the Deflated Sharpe Ratio (DSR). === Limitations === The main limitation of CPCV stems from its high computational cost. However, this cost can be managed by sampling a finite number of splits from the space of all possible combinations.
Read more →
Eduroam

eduroam (a portmanteau of education and roaming) is an international Wi-Fi internet access roaming service for users in research, higher education and further education. It provides researchers, teachers, and students network access when visiting an institution other than their own. Users are authenticated with credentials from their home institution, regardless of the location of the eduroam access point. Authorization to access the Internet and other resources are handled by the visited institution. Users do not have to pay to use eduroam. In some countries, Internet access via eduroam is also available at other locations than the participating institutions, e.g. in libraries, public buildings, railway stations, city centres and airports. It is also available at many primary and secondary education institutions in Brazil and the US. == History == The eduroam initiative started in 2002 when during the preparations for the creation of TERENA's task force TF-Mobility, Klaas Wierenga of SURFnet shared the idea of combining a RADIUS-based infrastructure with IEEE 802.1X technology to provide roaming network access across research and education networks. Initially, the service was joined by institutions in the Netherlands, Germany, Finland, Portugal, Croatia and the United Kingdom. Later, other NRENs in Europe embraced the idea and started joining the infrastructure, which was then called eduroam. Since 2004, the European Union co-funded further research and development work related to the eduroam service through the GN2 and GN3 projects. From September 2007, the European Union also funded through these projects the continued operation and maintenance of the eduroam service at the European level. The first non-European country to join eduroam was Australia, in December 2004. In Canada, eduroam started as an initiative of the University of British Columbia, which was later taken over by CANARIE as a service of its Canadian Access Federation. In the United States, eduroam was initially a pilot project between the National Science Foundation and the University of Tennessee (UTK). In 2012, Internet2 announced the addition of eduroam to its NET+ service offerings. AnyRoam LLC, a private company, was formed by former UTK staff to serve as an Internet2 active corporate member administering the US top-level servers. In 2021, Internet2 assumed direct management of the eduroam service for US-based organizations. == Technology == The eduroam service uses IEEE 802.1X as the authentication method and a hierarchical system of RADIUS servers. The hierarchy typically consists of RADIUS servers at the participating institutions, national RADIUS servers run by the National Roaming Operators, and regional top-level RADIUS servers for individual world regions. In some cases, institutions contact each other directly via DNS lookups () When a user visits a remote institution, the user's device presents their credentials to the local RADIUS server. That RADIUS server discovers that it is not responsible for the realm of the user's home institution and proxies the access request to another RADIUS server, typically the national RADIUS server. If the visited institution is in a different country than the home institution, the request is in turn proxied to the regional top-level RADIUS server, and then to the national RADIUS server of the user's home country. That national server forwards the credentials to the home institution, where they are verified. The RADIUS response travels back over the proxy-hierarchy to the visited institution and the user is granted access. In eduroam, the user credentials are always presented in the form of an EAP method (). The EAP method is responsible for ensuring that the users credentials are secure, and private. The users credentials can then travel via a number of intermediate servers, not under the control of the home institution of the user. This requirement limits the types of EAP methods that can be used. EAP methods which do not provide for security or privacy of user credentials cannot be used in eduroam. The most commonly used EAP methods in eduroam are EAP-TLS, PEAP, and EAP-TTLS. The methods used generally fall into two broad categories: those that use credentials in the form of some public-key mechanism with certificates and those that use so-called tunnelled authentication with "inner" passwords or other credentials. Most institutions use a tunnelled authentication method that requires a server certificate. These server certificates are used to set up a secure tunnel between the mobile device and the authentication server, through which the user credentials (e.g. name and password) are securely transported. A complication arises if the user's home institution does not use a two-letter country-code top-level domain as part of its realm, but a generic top-level domain such as .edu or .org. By inspection of such realms, it is not possible to determine which national RADIUS server the request should be routed to. Such domains will thus, by default, fail to work in international roaming. The workaround for this problem involves the creation of exceptions in the international RADIUS request routing tables; however, this workaround does not scale as the number of exception entries grows. Several solutions have been proposed to eliminate this workaround in the future, the most promising of which is RADIUS over TLS with Dynamic Discovery, which does not rely on static routing tables inside a RADIUS server configuration to route requests to their proper destination. Instead, the participating institution adds one NAPTR DNS resource record to its own domain's DNS zone, which states by which server eduroam authentication for the domain is handled. == Governance == GÉANT has established a lightweight global governance structure. Recognising the large variety in the organisation and funding of research and education (networking) in different countries and regions, rules imposed on the operations of eduroam are limited to technical and administrative requirements that are necessary to ensure the smooth and secure operations of eduroam worldwide. Moreover, the eduroam operators have the leading role in creating and maintaining the rules of the global eduroam governance. The Global eduroam Governance Committee (GeGC) has the central role in the global eduroam governance structure. While its structure has evolved over time, it presently has three representatives from each of five regions — mirroring those used by the Regional Internet registries — serving a two-year term. In addition, GÉANT may appoint one or more experts as non-voting members of the GeGC. == Geographical deployment == eduroam is available at selected locations in countries with a National Roaming Operator that has signed the eduroam Compliance Statement. Those sixty-seven countries are listed below. In addition, there may be pilot deployments in countries that are in the process of joining eduroam. === Middle East === eduroam is deployed in: === Europe === The NRENs that are members of the consortium of the GN3 project have joined the European eduroam confederation by signing the confederation's policy that requires its members to comply with a set of technical and organisational requirements, which are more specific than those in the global eduroam Compliance Statement. As a consequence, eduroam is deployed in the following countries: In addition, three NRENs that are associate members of the consortium of the GN3 project without voting rights joined the European eduroam confederation; they represent Belarus (UIIP), Moldova (RENAM) and Russia (Joint Supercomputer Center of the Russian Academy of Sciences). Finally, five NRENs not involved in the GN3 project joined the European eduroam confederation on a voluntary basis, enabling the deployment of the service in: The European top-level RADIUS servers are operated by SURFnet and Forskningsnettet. === Asia-Pacific === eduroam is deployed in the following countries and economies: The Asia-Pacific top-level RADIUS servers are operated by AARNet and by the University of Hong Kong. === North America === eduroam is deployed in: === Latin America === eduroam is deployed in: === Africa === eduroam is deployed in: The inter-African RADIUS servers are operated by West-African research and education network WACREN, the UbuntuNet Alliance and TENET.
Read more →
NYSERNet

NYSERNet, Inc. (New York State Education and Research Network), is a non-profit Internet service provider in New York State. It mainly provides Internet access to universities, colleges, museums, health care facilities, primary and secondary schools, and research institutions. == History == NYSERNet was founded in 1986 in Troy, New York. Its founders compared NYSERNet's network with the Erie Canal and considered it the next step in two centuries to draw the country together. NYSERNet's network reaches from Buffalo to New York City. Completed in 1987, it was the first statewide regional IP network in the United States.[1] Initial speed of 56 kbps was upgraded to T1 in 1989 and T3 in 1994. It was the original assignee of AS174 according to RFC1117. This ASN is used today by Cogent Communications for their global network.
Read more →
Data recovery

In computing, data recovery is a process of retrieving deleted, inaccessible, lost, corrupted, damaged, or overwritten data from secondary storage, removable media or files, when the data stored in them cannot be accessed in a usual way. The data is most often salvaged from storage media such as internal or external hard disk drives (HDDs), solid-state drives (SSDs), USB flash drives, magnetic tapes, CDs, DVDs, RAID subsystems, and other electronic devices. Recovery may be required due to physical damage to the storage devices or logical damage to the file system that prevents it from being mounted by the host operating system (OS). Logical failures occur when the hard drive devices are functional but the user or automated-OS cannot retrieve or access data stored on them. Logical failures can occur due to corruption of the engineering chip, lost partitions, firmware failure, or failures during formatting/re-installation. Data recovery can be a very simple or technical challenge. This is why there are specific software companies specialized in this field that help to get back data on your system. == About == The most common data recovery scenarios involve an operating system failure, malfunction of a storage device, logical failure of storage devices, accidental damage or deletion, etc. (typically, on a single-drive, single-partition, single-OS system), in which case the ultimate goal is simply to copy all important files from the damaged media to another new drive. This can be accomplished using a Live CD, or DVD by booting directly from a ROM or a USB drive instead of the corrupted drive in question. Many Live CDs or DVDs provide a means to mount the system drive and backup drives or removable media, and to move the files from the system drive to the backup media with a file manager or optical disc authoring software. Such cases can often be mitigated by disk partitioning and consistently storing valuable data files (or copies of them) on a different partition from the replaceable OS system files. Another scenario involves a drive-level failure, such as a compromised file system or drive partition, or a hard disk drive failure. In any of these cases, the data is not easily read from the media devices. Depending on the situation, solutions involve repairing the logical file system, partition table, or master boot record, or updating the firmware or drive recovery techniques ranging from software-based recovery of corrupted data, to hardware- and software-based recovery of damaged service areas (also known as the hard disk drive's "firmware"), to hardware replacement on a physically damaged drive which allows for the extraction of data to a new drive. If a drive recovery is necessary, the drive itself has typically failed permanently, and the focus is rather on a one-time recovery, salvaging whatever data can be read. In a third scenario, files have been accidentally "deleted" from a storage medium by the users. Typically, the contents of deleted files are not removed immediately from the physical drive; instead, references to them in the directory structure are removed, and thereafter space the deleted data occupy is made available for later data overwriting. In the mind of end users, deleted files cannot be discoverable through a standard file manager, but the deleted data still technically exists on the physical drive. In the meantime, the original file contents remain, often several disconnected fragments, and may be recoverable if not overwritten by other data files. The term "data recovery" is also used in the context of forensic applications or espionage, where data which have been encrypted, hidden, or deleted, rather than damaged, are recovered. Sometimes data present in the computer gets encrypted or hidden due to reasons like virus attacks which can only be recovered by some computer forensic experts. == Physical damage == A wide variety of failures can cause physical damage to storage media, which may result from human errors and natural disasters. CD-ROMs can have their metallic substrate or dye layer scratched off; hard disks can suffer from a multitude of mechanical failures, such as head crashes, PCB failure, and failed motors; tapes can simply break. Physical damage to a hard drive, even in cases where a head crash has occurred, does not necessarily mean permanent data loss. However, in extreme cases, such as prolonged exposure to moisture and corrosion —like the lost Bitcoin hard drive of James Howells, buried in the Newport landfill for over a decade — recovery is usually impossible. In rare cases, forensic techniques such as magnetic force microscopy (MFM) have been explored to detect residual magnetic traces when data holds exceptional value. Other techniques employed by many professional data recovery companies can typically salvage most, if not all, of the data that had been lost when the failure occurred. Of course, there are exceptions to this, such as cases where severe damage to the hard drive platters may have occurred. However, if the hard drive can be repaired and a full image or clone created, then the logical file structure can be rebuilt in most instances. Most physical damage cannot be repaired by end users. For example, opening a hard disk drive in a normal environment can allow airborne dust to settle on the platter and become caught between the platter and the read/write head. During normal operation, read/write heads float 3 to 6 nanometers above the platter surface, and the average dust particles found in a normal environment are typically around 30,000 nanometers in diameter. When these dust particles get caught between the read/write heads and the platter, they can cause new head crashes that further damage the platter and thus compromise the recovery process. Furthermore, end users generally do not have the hardware or technical expertise required to make these repairs. Consequently, data recovery companies are often employed to salvage important data with the more reputable ones using class 100 dust- and static-free cleanrooms. === Recovery techniques === Recovering data from physically damaged hardware can involve multiple techniques. Some damage can be repaired by replacing parts in the hard disk. This alone may make the disk usable, but there may still be logical damage. A specialized disk-imaging procedure is used to recover every readable bit from the surface. Once this image is acquired and saved on a reliable medium, the image can be safely analyzed for logical damage and will possibly allow much of the original file system to be reconstructed. ==== Hardware repair ==== A common misconception is that a damaged printed circuit board (PCB) may be simply replaced during recovery procedures by an identical PCB from a healthy drive. While this may work in rare circumstances on hard disk drives manufactured before 2003, it will not work on newer drives. Electronics boards of modern drives usually contain drive-specific adaptation data (generally a map of bad sectors and tuning parameters) and other information required to properly access data on the drive. Replacement boards often need this information to effectively recover all of the data. The replacement board may need to be reprogrammed. Some manufacturers (Seagate, for example) store this information on a serial EEPROM chip, which can be removed and transferred to the replacement board. Each hard disk drive has what is called a system area or service area; this portion of the drive, which is not directly accessible to the end user, usually contains drive's firmware and adaptive data that helps the drive operate within normal parameters. One function of the system area is to log defective sectors within the drive; essentially telling the drive where it can and cannot write data. The sector lists are also stored on various chips attached to the PCB, and they are unique to each hard disk drive. If the data on the PCB do not match what is stored on the platter, then the drive will not calibrate properly. In most cases the drive heads will click because they are unable to find the data matching what is stored on the PCB. == Logical damage == The term "logical damage" refers to situations in which the error is not a problem in the hardware and requires software-level solutions. === Corrupt partitions and file systems, media errors === In some cases, data on a hard disk drive can be unreadable due to damage to the partition table or file system, or to (intermittent) media errors. In the majority of these cases, at least a portion of the original data can be recovered by repairing the damaged partition table or file system using specialized data recovery software such as TestDisk; software like ddrescue can image media despite intermittent errors, and image raw data when there is partition table or file system damage. This type of data recovery can be performed by people without expertise in drive hardware as it requires no special physica
Read more →
CinePlayer

CinePlayer is a software based media player used to review Digital Cinema Packages (DCP) without the need for a digital cinema server by Doremi Labs. CinePlayer can play back any DCP, not just those created by Doremi Mastering products. In addition to playing DCPs, CinePlayer can also playback JPEG2000 image sequences and many popular multimedia file types. There are two versions of CinePlayer available, standard and Pro. The standard version supports playback of non-encrypted, 2D DCP's up to 2K resolution. The Pro version supports playback of encrypted, 2D or 3D DCP's with subtitles up to 4K resolution. == Supported formats == === Containers === AVI MOV MXF MPG TS WMV M2TS MTS MP4 MKV === Video codecs === JPEG2000 ProRes 422 DNxHD YUV Uncompressed 8-10 bits DIVX XVID MPEG4 AVC / H-264 VC-1 MPEG2 === Supported image sequences === BMP TIFF TGA DPX JPG J2C === Supported audio files === WAV MP3 WMA MP2
Read more →
Strategic Air Command Digital Information Network

The Strategic Air Command DIgital Network (SACDIN) was a United States military computer network that provided computerized record communications, replacing the Data Transmission Subsystem and part of the Data Display Subsystem of the SAC Automated Command and Control System. SACDIN enabled a rapid flow of communications from headquarters SAC to its fielded forces, such as B-52 bases and ICBM Launch Control Centers. == Logistics == Major portions of SACDIN were developed, engineered and installed by the International Telephone and Telegraph (ITT) company, under contract to the Electronic Systems Center. == Chronology == 1969 - Headquarters SAC submits a request to the Joint Chiefs of Staff to study an expanded communications system, known as the SAC Total Information Network (SATIN). It would interconnect Air Force Satellite Communications (AFSATCOM), Advanced Airborne Command Post (AABNCP), Airborne Command Post (ABNCP), high frequency/single sideband radio HF/SSB radio, SAC Automated Command and Control System (SACCS), Automatic Digital Information Network (AUTODIN), Survivable Low Frequency Communications System (SLFCS) and Command Data Buffer (CDB) 1977 1 November - SATIN IV was effectively terminated by Congress. The restructured program was renamed SAC Digital Network (SACDIN), and was formulated to meet SAC's minimum essential data communications requirements, but also had the capability to grow in a modular fashion. 1986 ?? ??? - SACDIN replaces much of the SAC Automated Command and Control System (SACCS) and the SAC Automated Total Information Network (SATIN)
Read more →
Star Awards for Social Media Award

The Star Awards for Social Media Award was an award presented annually from 2014 to 2016 at the Star Awards, where Mediacorp of Singapore recognises entertainers under their employment with awards for artistic and technical merit for outstanding performances of the year. == History == The category was introduced in 2014, at the 20th Star Awards ceremony; Jeanette Aw received the award and it is given in honour of a Mediacorp artiste with the most social media engagement. The results are based on the calculations from three international social media analysis systems; artistes must be active on at least one of the following platforms in order to qualify: Facebook, Twitter and Instagram. Since its inception, the award has been given to two artistes. Carrie Wong is the most recent and final winner in this category. Since the ceremony held in 2016, Aw remains as the only artiste to win in this category twice, surpassing Wong who has one win. The award was discontinued from 2017 onwards as the popularity element of the award is already represented in the Top 10 Most Popular Male Artistes and Top 10 Most Popular Female Artistes awards. == Recipients ==
Read more →
Polygraphic substitution

Polygraphic substitution is a substitution cipher in which a uniform substitution is performed on blocks of letters. When the length of the block is specifically known, more precise terms are used: for instance, a cipher in which pairs of letters are substituted is bigraphic. As a concept, polygraphic substitution contrasts with monoalphabetic (or simple) substitutions in which individual letters are uniformly substituted, or polyalphabetic substitutions in which individual letters are substituted in different ways depending on their position in the text. In theory, there is some overlap in these definitions; one could conceivably consider a Vigenère cipher with an eight-letter key to be an octographic substitution. In practice, this is not a useful observation since it is far more fruitful to consider it to be a polyalphabetic substitution cipher. == Specific ciphers == In 1563, Giambattista della Porta devised the first bigraphic substitution. However, it was nothing more than a matrix of symbols. In practice, it would have been all but impossible to memorize, and carrying around the table would lead to risks of falling into enemy hands. In 1854, Charles Wheatstone came up with the Playfair cipher, a keyword-based system that could be performed on paper in the field. This was followed up over the next fifty years with the closely related four-square and two-square ciphers, which are slightly more cumbersome but offer slightly better security. In 1929, Lester S. Hill developed the Hill cipher, which uses matrix algebra to encrypt blocks of any desired length. However, encryption is very difficult to perform by hand for any sufficiently large block size, although it has been implemented by machine or computer. This is therefore on the frontier between classical and modern cryptography. == Cryptanalysis of general polygraphic substitutions == Polygraphic systems do provide a significant improvement in security over monoalphabetic substitutions. Given an individual letter 'E' in a message, it could be encrypted using any of 52 instructions depending on its location and neighbors, which can be used to great advantage to mask the frequency of individual letters. However, the security boost is limited; while it generally requires a larger sample of text to crack, it can still be done by hand. One can identify a polygraphically-encrypted text by performing a frequency chart of polygrams and not merely of individual letters. These can be compared to the frequency of plaintext English. The distribution of digrams is even more stark than individual letters. For example, the six most common letters in English (23%) represent approximately half of English plaintext, but it takes only the most frequent 8% of the 676 digrams to achieve the same potency. In addition, even in a plaintext many thousands of characters long, one would expect that nearly half of the digrams would not occur, or only barely. In addition, looking over the text one would expect to see a fairly regular scattering of repeated text in multiples of the block length and relatively few that are not multiples. Cracking a code identified as polygraphic is similar to cracking a general monoalphabetic substitution except with a larger 'alphabet'. One identifies the most frequent polygrams, experiments with replacing them with common plaintext polygrams, and attempts to build up common words, phrases, and finally meaning. Naturally, if the investigation led the cryptanalyst to suspect that a code was of a specific type, like a Playfair or order-2 Hill cipher, then they could use a more specific attack.
Read more →
Exploration–exploitation dilemma

The exploration–exploitation dilemma, also known as the explore–exploit tradeoff, is a fundamental concept in decision-making that arises in many domains. It is depicted as the balancing act between two opposing strategies. Exploitation involves choosing the best option based on current knowledge of the system (which may be incomplete or misleading), while exploration involves trying out new options that may lead to better outcomes in the future at the expense of an exploitation opportunity. Finding the optimal balance between these two strategies is a crucial challenge in many decision-making problems whose goal is to maximize long-term benefits. == Application in machine learning == In the context of machine learning, the exploration–exploitation tradeoff is fundamental in reinforcement learning (RL), a type of machine learning that involves training agents to make decisions based on feedback from the environment. Crucially, this feedback may be incomplete or delayed. The agent must decide whether to exploit the current best-known policy or explore new policies to improve its performance. === Multi-armed bandit methods === The multi-armed bandit (MAB) problem was a classic example of the tradeoff, and many methods were developed for it, such as epsilon-greedy, Thompson sampling, and the upper confidence bound (UCB). See the page on MAB for details. In more complex RL situations than the MAB problem, the agent can treat each choice as a MAB, where the payoff is the expected future reward. For example, if the agent performs an epsilon-greedy method, then the agent will often "pull the best lever" by picking the action that had the best predicted expected reward (exploit). However, it would pick a random action with probability epsilon (explore). Monte Carlo tree search, for example, uses a variant of the UCB method. === Exploration problems === There are some problems that make exploration difficult. Sparse reward. If rewards occur only once a long while, then the agent might not persist in exploring. Furthermore, if the space of actions is large, then the sparse reward would mean the agent would not be guided by the reward to find a good direction for deeper exploration. A standard example is Montezuma's Revenge. Deceptive reward. If some early actions give immediate small reward, but other actions give later large reward, then the agent might be lured away from exploring the other actions. Noisy TV problem. If certain observations are irreducibly noisy (such as a television showing random images), then the agent might be trapped exploring those observations (watching the television). === Exploration reward === This section based on. The exploration reward (also called exploration bonus) methods convert the exploration-exploitation dilemma into a balance of exploitations. That is, instead of trying to get the agent to balance exploration and exploitation, exploration is simply treated as another form of exploitation, and the agent simply attempts to maximize the sum of rewards from exploration and exploitation. The exploration reward can be treated as a form of intrinsic reward. We write these as r t i , r t e {\displaystyle r_{t}^{i},r_{t}^{e}} , meaning the intrinsic and extrinsic rewards at time step t {\displaystyle t} . However, exploration reward is different from exploitation in two regards: The reward of exploitation is not freely chosen, but given by the environment, but the reward of exploration may be picked freely. Indeed, there are many different ways to design r t i {\displaystyle r_{t}^{i}} described below. The reward of exploitation is usually stationary (i.e. the same action in the same state gives the same reward), but the reward of exploration is non-stationary (i.e. the same action in the same state should give less and less reward). Count-based exploration uses N n ( s ) {\displaystyle N_{n}(s)} , the number of visits to a state s {\displaystyle s} during the time-steps 1 : n {\displaystyle 1:n} , to calculate the exploration reward. This is only possible in small and discrete state space. Density-based exploration extends count-based exploration by using a density model ρ n ( s ) {\displaystyle \rho _{n}(s)} . The idea is that, if a state has been visited, then nearby states are also partly-visited. In maximum entropy exploration, the entropy of the agent's policy π {\displaystyle \pi } is included as a term in the intrinsic reward. That is, r t i = − ∑ a π ( a | s t ) ln ⁡ π ( a | s t ) + ⋯ {\displaystyle r_{t}^{i}=-\sum _{a}\pi (a|s_{t})\ln \pi (a|s_{t})+\cdots } . === Prediction-based === This section based on. The forward dynamics model is a function for predicting the next state based on the current state and the current action: f : ( s t , a t ) ↦ s t + 1 {\displaystyle f:(s_{t},a_{t})\mapsto s_{t+1}} . The forward dynamics model is trained as the agent plays. The model becomes better at predicting state transition for state-action pairs that had been done many times. A forward dynamics model can define an exploration reward by r t i = ‖ f ( s t , a t ) − s t + 1 ‖ 2 2 {\displaystyle r_{t}^{i}=\|f(s_{t},a_{t})-s_{t+1}\|_{2}^{2}} . That is, the reward is the squared-error of the prediction compared to reality. This rewards the agent to perform state-action pairs that had not been done many times. This is however susceptible to the noisy TV problem. Dynamics model can be run in latent space. That is, r t i = ‖ f ( s t , a t ) − ϕ ( s t + 1 ) ‖ 2 2 {\displaystyle r_{t}^{i}=\|f(s_{t},a_{t})-\phi (s_{t+1})\|_{2}^{2}} for some featurizer ϕ {\displaystyle \phi } . The featurizer can be the identity function (i.e. ϕ ( x ) = x {\displaystyle \phi (x)=x} ), randomly generated, the encoder-half of a variational autoencoder, etc. A good featurizer improves forward dynamics exploration. The Intrinsic Curiosity Module (ICM) method trains simultaneously a forward dynamics model and a featurizer. The featurizer is trained by an inverse dynamics model, which is a function for predicting the current action based on the features of the current and the next state: g : ( ϕ ( s t ) , ϕ ( s t + 1 ) ) ↦ a t {\displaystyle g:(\phi (s_{t}),\phi (s_{t+1}))\mapsto a_{t}} . By optimizing the inverse dynamics, both the inverse dynamics model and the featurizer are improved. Then, the improved featurizer improves the forward dynamics model, which improves the exploration of the agent. Random Network Distillation (RND) method attempts to solve this problem by teacher–student distillation. Instead of a forward dynamics model, it has two models f , f ′ {\displaystyle f,f'} . The f ′ {\displaystyle f'} teacher model is fixed, and the f {\displaystyle f} student model is trained to minimize ‖ f ( s ) − f ′ ( s ) ‖ 2 2 {\displaystyle \|f(s)-f'(s)\|_{2}^{2}} on states s {\displaystyle s} . As a state is visited more and more, the student network becomes better at predicting the teacher. Meanwhile, the prediction error is also an exploration reward for the agent, and so the agent learns to perform actions that result in higher prediction error. Thus, we have a student network attempting to minimize the prediction error, while the agent attempting to maximize it, resulting in exploration. The states are normalized by subtracting a running average and dividing a running variance, which is necessary since the teacher model is frozen. The rewards are normalized by dividing with a running variance. Exploration by disagreement trains an ensemble of forward dynamics models, each on a random subset of all ( s t , a t , s t + 1 ) {\displaystyle (s_{t},a_{t},s_{t+1})} tuples. The exploration reward is the variance of the models' predictions. === Noise === For neural network–based agents, the NoisyNet method changes some of its neural network modules by noisy versions. That is, some network parameters are random variables from a probability distribution. The parameters of the distribution are themselves learnable. For example, in a linear layer y = W x + b {\displaystyle y=Wx+b} , both W , b {\displaystyle W,b} are sampled from Gaussian distributions N ( μ W , Σ W ) , N ( μ b , Σ b ) {\displaystyle {\mathcal {N}}(\mu _{W},\Sigma _{W}),{\mathcal {N}}(\mu _{b},\Sigma _{b})} at every step, and the parameters μ W , Σ W , μ b , Σ b {\displaystyle \mu _{W},\Sigma _{W},\mu _{b},\Sigma _{b}} are learned via the reparameterization trick.
Read more →
Social media surgery

A social media surgery is a gathering at which volunteer "surgeons" with expertise in using web tools, chiefly social media, offer free advice in using such tools, to representatives ("patients") of non-profit organisations, charities, community groups and activists, with "no boring speeches or jargon". The idea was conceived by Pete Ashton, with Nick Booth of Podnosh Ltd, who ran the first such surgery in Birmingham, England, on 15 October 2008. In July 2009, a spin-off surgery (dubbed the "Social media mob") started in Mosman, Australia, and in January 2010, the first spin-off surgery in Africa was held. On 16 February 2012, it was announced that the Social Media Surgery movement had won "the Prime Minister’s Big Society Award". Prime Minister David Cameron said: This is an excellent initiative - such a simple idea and yet so effective. The popularity of these surgeries and the fact that they have inspired so many others across the country to follow in their footsteps, is testament to its brilliance. Congratulations to Nick and all the volunteers who have shared their time and expertise to help so many local groups make the most of the internet to support their community. A great example of the Big Society in action. The scheme also won the 2013 Adult Learners' Week "BBC Learning Through Technology Award".
Read more →
Perfectly Imperfect (platform)

Perfectly Imperfect is an online newsletter and social media platform. It was initially founded in 2020 as a biweekly email newsletter that focused on recommendations. In January 2024, Perfectly Imperfect launched PI.FYI, a social media platform. The platform is based around sharing recommendations. Its main feed is presented in reverse chronological order and is not algorithmically curated. == History == Perfectly Imperfect was started during the COVID-19 pandemic by Tyler Bainbridge, alongside college friends Alex Cushing and Serey Morm, whom he met at UMass Lowell; Morm later departed. Motivated by a dissatisfaction with algorithm-driven recommendation culture, they launched on Substack in September 2020. Its early newsletter format, PI, published brief recommendation lists and personal notes from contributors. Contributors have included a mix of underground artists and more established creative figures, such as Charli XCX, Chloe Cherry, Chloe Wise, and Meetka Otto. In October 2024, PI announced it was leaving Substack to launch its own site. == Overview == The current platform, PI.FYI, features both editorial content (guest columns, long-form essays, staff picks) and user-generated recommendations. The platform also supports "Ask" posts, where users can solicit recommendations from the community, and allows commenting, liking, and profile customization. In August 2025, it launched an events feature. In 2022, Perfectly Imperfect hosted their first offline event at Baby's All Right in Brooklyn, with a performance by The Dare. They have since expanded their event promotion/sponsorship to markets such as Los Angeles, San Francisco, and even Auckland.
Read more →